Feasibility study on quantification and authentication of the cassava starch content in wheat flour for bread-making using NIR spectroscopy and digital images.

This works proposed a feasibility study on NIR spectroscopy and chemometrics-assisted color histogram-based analytical systems (CACHAS) to determine and authenticate the cassava starch content in wheat flour. Prediction results of partial least squares (PLS) achieved coefficient of correlation (rpred) of 0.977 and root mean square error of prediction (RMSEP) of 1.826 mg kg-1 for the certified additive-free wheat flour, while rpred of 0.995 and RMSEP of 1.004 mg kg-1 were obtained for the commercial wheat flour containing chemical additives. Additionally, Data-Driven Soft Independent Modelling of Class Analogy (dd-SIMCA) presented similar predictive ability using NIR and CACHAS for the certified wheat flour, authenticating all target samples, besides correctly recognizing samples that could represent a fraud. No satisfactory results were obtained for the commercial wheat flour. Therefore, NIR spectroscopy is more useful to offer definitive quantitative and qualitative analysis, while CACHAS can only provide an alternative preliminary analysis.

Scores selection via Fisher's discriminant power in PCA-LDA to improve the classification of food data.

This paper proposes an adaptation of the Fisher's discriminability criterion (named here as discriminant power, DP) for choosing principal components (obtained from Principal Component Analysis, PCA), which will be used to construct supervised Linear Discriminant Analysis (LDA) models for solving classification problems of food data. The proposed PCA-DP-LDA algorithm was then applied to (i) simulated data, (ii) classify soybean oils with respect to expiration date, and (iii) identify cachaça adulteration with wood extracts that simulated aging. For comparison, PCA-DP-LDA was evaluated against conventional PCA-LDA (based on explained variance) and Partial Least Squares-Discriminant Analysis (PLS-DA). Among them, PCA-DP-LDA achieved the most parsimonious and interpretable results, with similar or better classification performance. Therefore, the new algorithm can be considered a good alternative to the already well-established discriminant methods, being potentially applied where the discriminability of the principal components may not follow the same behavior of the explained variance.

Simultaneous identification of the wood types in aged cachaças and their adulterations with wood extracts using digital images and SPA-LDA.

Cachaça is a sugarcane-derived alcoholic spirit exclusively produced in Brazil. It can be aged in barrels made from different types of wood, similar to other distilled beverages. The choice of wood type promotes different effects on color, flavor, aroma and consequently the price of cachaça, favoring fraudulent activities. This paper proposes the simultaneous identification of different wood types in aged cachaças and their adulterations with wood extracts using a digital-image based methodology employing color histograms obtained from digital images associated with pattern recognition methods, without any sample preparation step. Linear Discriminant Analysis, coupled with Successive Projections Algorithm for variable selection (SPA-LDA), obtained the best results, reaching accuracy, sensitivity, and specificity rates higher than 90.0% in the test set. This can be a rapid and reliable tool to prevent fraudulent labeling; ensuring that what is on the label reflects the quality of aged cachaças, affording security to consumers and regulatory agencies.

Vis-NIR spectrometric determination of Brix and sucrose in sugar production samples using kernel partial least squares with interval selection based on the successive projections algorithm.

This paper proposes a new variable selection method for nonlinear multivariate calibration, combining the Successive Projections Algorithm for interval selection (iSPA) with the Kernel Partial Least Squares (Kernel-PLS) modelling technique. The proposed iSPA-Kernel-PLS algorithm is employed in a case study involving a Vis-NIR spectrometric dataset with complex nonlinear features. The analytical problem consists of determining Brix and sucrose content in samples from a sugar production system, on the basis of transflectance spectra. As compared to full-spectrum Kernel-PLS, the iSPA-Kernel-PLS models involve a smaller number of variables and display statistically significant superiority in terms of accuracy and/or bias in the predictions.

Using near infrared spectroscopy to classify soybean oil according to expiration date.

A rapid and non-destructive methodology is proposed for the screening of edible vegetable oils according to conservation state expiration date employing near infrared (NIR) spectroscopy and chemometric tools. A total of fifty samples of soybean vegetable oil, of different brands andlots, were used in this study; these included thirty expired and twenty non-expired samples. The oil oxidation was measured by peroxide index. NIR spectra were employed in raw form and preprocessed by offset baseline correction and Savitzky-Golay derivative procedure, followed by PCA exploratory analysis, which showed that NIR spectra would be suitable for the classification task of soybean oil samples. The classification models were based in SPA-LDA (Linear Discriminant Analysis coupled with Successive Projection Algorithm) and PLS-DA (Discriminant Analysis by Partial Least Squares). The set of samples (50) was partitioned into two groups of training (35 samples: 15 non-expired and 20 expired) and test samples (15 samples 5 non-expired and 10 expired) using sample-selection approaches: (i) Kennard-Stone, (ii) Duplex, and (iii) Random, in order to evaluate the robustness of the models. The obtained results for the independent test set (in terms of correct classification rate) were 96% and 98% for SPA-LDA and PLS-DA, respectively, indicating that the NIR spectra can be used as an alternative to evaluate the degree of oxidation of soybean oil samples.

Digital image-based classification of biodiesel.

This work proposes a simple, rapid, inexpensive, and non-destructive methodology based on digital images and pattern recognition techniques for classification of biodiesel according to oil type (cottonseed, sunflower, corn, or soybean). For this, differing color histograms in RGB (extracted from digital images), HSI, Grayscale channels, and their combinations were used as analytical information, which was then statistically evaluated using Soft Independent Modeling by Class Analogy (SIMCA), Partial Least Squares Discriminant Analysis (PLS-DA), and variable selection using the Successive Projections Algorithm associated with Linear Discriminant Analysis (SPA-LDA). Despite good performances by the SIMCA and PLS-DA classification models, SPA-LDA provided better results (up to 95% for all approaches) in terms of accuracy, sensitivity, and specificity for both the training and test sets. The variables selected Successive Projections Algorithm clearly contained the information necessary for biodiesel type classification. This is important since a product may exhibit different properties, depending on the feedstock used. Such variations directly influence the quality, and consequently the price. Moreover, intrinsic advantages such as quick analysis, requiring no reagents, and a noteworthy reduction (the avoidance of chemical characterization) of waste generation, all contribute towards the primary objective of green chemistry.

Using color histograms and SPA-LDA to classify bacteria.

In this work, a new approach is proposed to verify the differentiating characteristics of five bacteria (Escherichia coli, Enterococcus faecalis, Streptococcus salivarius, Streptococcus oralis, and Staphylococcus aureus) by using digital images obtained with a simple webcam and variable selection by the Successive Projections Algorithm associated with Linear Discriminant Analysis (SPA-LDA). In this sense, color histograms in the red-green-blue (RGB), hue-saturation-value (HSV), and grayscale channels and their combinations were used as input data, and statistically evaluated by using different multivariate classifiers (Soft Independent Modeling by Class Analogy (SIMCA), Principal Component Analysis-Linear Discriminant Analysis (PCA-LDA), Partial Least Squares Discriminant Analysis (PLS-DA) and Successive Projections Algorithm-Linear Discriminant Analysis (SPA-LDA)). The bacteria strains were cultivated in a nutritive blood agar base layer for 24 h by following the Brazilian Pharmacopoeia, maintaining the status of cell growth and the nature of nutrient solutions under the same conditions. The best result in classification was obtained by using RGB and SPA-LDA, which reached 94 and 100 % of classification accuracy in the training and test sets, respectively. This result is extremely positive from the viewpoint of routine clinical analyses, because it avoids bacterial identification based on phenotypic identification of the causative organism using Gram staining, culture, and biochemical proofs. Therefore, the proposed method presents inherent advantages, promoting a simpler, faster, and low-cost alternative for bacterial identification.